Regulation of biosynthesis of the basic fibroblast growth factor binding domains of heparan sulfate by heparan sulfate-N-deacetylase/N-sulfotransferase expression

Heparan sulfate-N-deacetylase/N-sulfotransferase catalyzes both the N-deacetylation and N-sulfation reactions that initiate the modification of the oligosaccharide backbone of heparan sulfate (HS). The glycosaminoglycan polymer appears to modulate the activity of growth factors by mediating their initial binding. To understand how the biosynthesis of these binding sites is regulated, a rat liver-derived cDNA encoding the above activities was overexpressed in a COS cell mutant (CM-15) that has reduced levels of the enzyme and binds poorly to immobilized basic fibroblast growth factor (bFGF). This resulted in increased synthesis of sulfated blocks of decasaccharide size or longer. These blocks exhibited high affinity binding to bFGF and contained a high content of 2-O-sulfated iduronate and at least five consecutive N-sulfated disaccharides. An increase in the synthesis of these high affinity blocks was not seen in transfected wild-type COS cells even though they showed a 4-fold increase of both enzyme activities, suggesting that once sufficient levels of highly sulfated blocks of saccharides with high affinity for bFGF are attained, no further synthesis of these domains occurs.

Guanosine diphosphatase is required for protein and sphingolipid glycosylation in the Golgi lumen of Saccharomyces cerevisiae

Current models for nucleotide sugar use in the Golgi apparatus predict a critical role for the lumenal nucleoside diphosphatase. After transfer of sugars to endogenous macromolecular acceptors, the enzyme converts nucleoside diphosphates to nucleoside monophosphates which in turn exit the Golgi lumen in a coupled antiporter reaction, allowing entry of additional nucleotide sugar from the cytosol. To test this model, we cloned the gene for the S. cerevisiae guanosine diphosphatase and constructed a null mutation. This mutation should reduce the concentrations of GDP-mannose and GMP and increase the concentration of GDP in the Golgi lumen. The alterations should in turn decrease mannosylation of proteins and lipids in this compartment. In fact, we found a partial block in O- and N-glycosylation of proteins such as chitinase and carboxypeptidase Y and underglycosylation of invertase. In addition, mannosylinositolphosphorylceramide levels were drastically reduced.

A single protein catalyzes both N-deacetylation and N-sulfation during the biosynthesis of heparan sulfate

Heparan sulfate is a highly sulfated carbohydrate polymer that binds to and modulates the activities of numerous proteins. The formation of these protein-binding domains in heparan sulfate is dependent on a series of biosynthetic reactions that modify the polysaccharide backbone; the initiating and rate-limiting steps of this process are the N-deacetylation and N-sulfation of N-acetylglucosamine residues in the polymer. We now report that in the rat liver, biosynthesis of heparan sulfate utilizes a single protein that possesses both N-deacetylase and N-sulfotransferase activities. This was accomplished by demonstrating that both activities resided in a purified soluble fusion protein containing the Golgi-lumenal portion of the enzyme. We propose that this protein be renamed the rat liver Golgi heparan sulfate N-deacetylase/N-sulfotransferase.

Molecular cloning and expression of rat liver N-heparan sulfate sulfotransferase

N-Heparan sulfate sulfotransferase catalyzes the transfer of sulfate from 3'-phosphoadenosine 5'-phosphosulfate to the nitrogen of glucosamine in heparan sulfate. The enzyme has been previously purified to apparent homogeneity from rat liver (Brandan, E., and Hirschberg, C. B. (1988) J. Biol. Chem. 263, 2417-2422). We have now cloned the rat liver enzyme using the following strategy: (a) the amino acid sequence was obtained from tryptic peptides of the purified protein, (b) mixed oligonucleotides were generated based on the sequence of the tryptic peptides, (c) a polymerase chain reaction fragment was obtained using mixed oligonucleotide interprimer amplification of cDNA, and (d) this fragment was used to screen rat liver lambda gt 10 and lambda ZAP libraries. Three clones were obtained, one of which seems to contain the complete coding sequence of the N-heparan sulfate sulfotransferase (N-HSST). Evidence that the cDNA clone corresponds to the previously purified and characterized N-HSST was the following: (a) the predicted sequence of the N-HSST contains all of the 11 tryptic peptides obtained from the purified protein, (b) when a cDNA containing the sequence coding for the N-HSST was introduced in a eukaryotic expression vector and transfected in COS-1 cells, the enzyme activity was expressed 9-fold over controls, and (c) the characteristic of the predicted protein fits with the purified protein in terms of molecular weight, membrane localization, and its being an N-linked glycoprotein. The size of the longest cDNA isolated is 4.1 kilobases, which is in close agreement with the 4.2-kilobase size of one of the mRNA observed in Northern analyses. In addition, messages of 7.0 and 8.5 kilobases were also observed, suggesting that a large portion is untranslated. The latter messages were the major mRNA species detected.

Translocation of ATP into the lumen of rough endoplasmic reticulum-derived vesicles and its binding to luminal proteins including BiP (GRP 78) and GRP 94

Rat liver and canine pancreas rough endoplasmic reticulum-derived vesicles, which were sealed and of the same topographical orientation as in vivo, were used in a system in vitro to demonstrate translocation of ATP into their lumen. Translocation of ATP is saturable (apparent Km: 3-4 microM and Vmax: 3-7 pmol/min/mg of protein) and protein mediated because treatment of intact vesicles with Pronase, N-ethylmaleimide, or 4,4'-diisothiocyanostilbene-2,2'-disulfonic acid inhibit transport. The entire ATP molecule is being translocated; this was shown by high performance liquid chromatography analysis and the use of a nonhydrolyzable analog. Control experiments rule out that translocation of ATP attributed to rough endoplasmic reticulum-derived vesicles is due to contamination by mitochondria and Golgi vesicles. Following translocation of ATP into the lumen of the vesicles, binding to luminal proteins including BiP (immunoglobulin heavy chain-binding protein-glucose-regulated protein 78) and glucose-regulated protein 94 was observed. This binding appeared to be specific because similar experiments with GTP were negative. These studies strongly suggest that translocation of ATP into the rough endoplasmic reticulum lumen may serve as a mechanism for making ATP available in proposed energy requiring reactions within the lumen.

Reconstitution into proteoliposomes and partial purification of the Golgi apparatus membrane UDP-galactose, UDP-xylose, and UDP-glucuronic acid transport activities

Previous studies in vitro on proteoglycan biosynthesis from our laboratory have shown that nucleotide sugar precursors of all the sugars of the linkage oligosaccharides (xylose, galactose, and glucuronic acid) and of the glycosaminoglycans (N-acetylglucosamine, N-galactosamine, and glucuronic acid) are transported by specific carriers into the lumen of Golgi vesicles. More recently, we also reported the reconstitution in phosphatidylcholine liposomes of detergent-solubilized Golgi membrane proteins containing transport activities of CMP-sialic acid and adenosine-3'-phosphate-5'-phosphosulfate. We have now completed the successful reconstitution into liposomes of the Golgi membrane transport activities of UDP-galactose, UDP-xylose, and UDP-glucuronic acid. Transport of these nucleotide sugars into Golgi protein proteoliposomes occurred with the same affinity, temperature dependence, and sensitivity to inhibitors as observed with intact Golgi vesicles. Preloading of proteoliposomes with UMP, the putative antiporter for Golgi vesicle transport of these nucleotide sugars, stimulated transport of the nucleotide sugars by 2-3-fold. Transport of UDP-xylose into Golgi protein proteoliposomes was dependent on the presence of endogenous Golgi membrane lipids while that of UDP-galactose and UDP-glucuronic acid was not. This suggests a possible stabilizing or regulatory role for Golgi lipids on the UDP-xylose translocator. Finally, we have also shown that detergent-solubilized Golgi membrane translocator proteins can be partially purified by an ion-exchange chromatographic step before successful reconstitution into liposomes, demonstrating that this reconstitution approach can be used for the biochemical purification of these transporters.

Topography of glycosylation reactions in the endoplasmic reticulum

A variety of distinct protein glycosylation reactions occur in the endoplasmic reticulum (ER) of eukaryotic cells. In some instances, both the proteins to be glycosylated and the precursor sugar donors must be translocated across the membrane from the cytoplasm to the lumen of the ER. Elucidation of the individual steps in each of the glycosylation pathways has revealed the topographic complexity of these reactions.

A guanosine diphosphatase enriched in Golgi vesicles of Saccharomyces cerevisiae. Purification and characterization

We have recently described a luminal guanosine diphosphatase activity in Golgi-like vesicles of Saccharomyces cerevisiae (Abeijon, C., Orlean, P., Robbins, P. W., and Hirschberg, C. B. (1989) Proc. Natl. Acad. Sci. U. S. A. 86, 6935-6939). The presumed in vivo role of this enzyme is to convert GDP into GMP. GDP is a reaction product following outer-chain mannosylation of luminal proteins and a known inhibitor of mannosyltransferases. It is hypothesized that GMP then returns to the cytosol. We have purified this enzyme to apparent homogeneity. Following solubilization from a membrane pellet using a buffer containing Triton X-100, the enzyme was purified on a concanavalin A-Sepharose column followed by Mono Q fast protein liquid chromatography (FPLC) and Superose-12 FPLC columns. After treatment with endoglycosidase H, the deglycosylated active enzyme was applied to a second Mono Q FPLC column and a phenyl-Superose FPLC column. The final enzyme activity was enriched 6500-fold over that of the Triton X-100 extract. The apparant molecular mass of the deglycosylated enzyme is 47 kDa. The purified enzyme is highly specific for guanosine diphosphate, requires Ca2+ for maximal activity, and has a broad pH optimum between 7.4 and 8.2. The apparent Km for GDP is 0.1 mM; the Vmax is 4.9 mmol/min/mg of protein. An enzyme activity with similar substrate specificity has also been detected in membranes of Schizosaccharomyces pombe.

Topography of initiation of N-glycosylation reactions

Previous studies on the topography of the reactions leading to the formation of dolichol-P-P-Glc-NAc2Man9Glc3 have shown that these occur on both sides of the endoplasmic reticulum membrane (Hirschberg, C. B., and Snider, M. D. (1987) Annu. Rev. Biochem. 56, 63-87). Dolichol-P-P-GlcNAc2Man5 has been detected on the cytoplasmic side of the endoplasmic reticulum membrane while the subsequent dolichol-oligosaccharide intermediates face the lumen. Less clear is the side of the membrane where dolichol-P-P-GlcNAc2 is assembled. We now present evidence strongly suggesting that the active sites of the enzymes catalyzing the synthesis of this latter intermediate are on the cytoplasmic side of the endoplasmic reticulum membrane. In addition, dolichol-P-P-GlcNAc2 has also been detected on this side. Incubations of sealed, "right side out" rat liver endoplasmic reticulum-derived vesicles with [beta-32P] UDP-GlcNAc in the presence of 5-Br-UMP resulted in the formation of radiolabeled dolichol-P-P-GlcNAc and dolichol-P-P-GlcNAc2 under conditions where there was complete inhibition of transport of the nucleotide sugar. In other experiments with the above radiolabeled nucleotide sugar and sealed vesicles, it was demonstrated that EDTA (a membrane-impermeable reagent) inhibited the N-acetylglucosamine-1-phosphate transferase under conditions where transport of the nucleotide sugar into the lumen was unaffected. Finally, sealed vesicles were first incubated with [32P]UDP-GlcNAc and subsequently with UDP-Gal and soluble galactosyltransferase. This resulted in galactosylation of dolichol-P-P-GlcNAc2. The above results, together with the previous observations, strongly suggest that all reactions leading to this latter dolichol intermediate occur on the cytosolic side of the endoplasmic reticulum membrane.

Differential association and distribution of acetyl- and butyrylcholinesterases within rat liver subcellular organelles

Rat liver cholinesterases were found to share properties and characteristics with those expressed in cholinergic tissues. The distribution and presence of different molecular forms of cholinesterases in different subcellular organelles of rat liver were studied. The rough and smooth endoplasmic reticulum and Golgi apparatus were enriched in the G4 molecular form of acetylcholinesterase (AChE) (relative to the G2 molecular form), while the inverse was found in the plasma membrane. The interaction of these molecular forms of AChE with the Golgi membrane was studied in detail. Approximately one-half of the G4 form was free within the lumen while the remainder was an intrinsic membrane protein; all the G2 molecular form was anchored to the membrane via phosphatidylinositol. Only the G1 and G2 molecular forms of butyrylcholinesterase (BuChE) were found in the above subcellular organelles; both molecular forms were soluble within the lumen of Golgi vesicles. These results indicate that rat liver expresses several molecular forms of AChE which have multiple interactions with membranes and that liver is unlikely to be the source of the G4 form of BuChE present in high concentration in the plasma.

Topography of glycosylation in yeast: characterization of GDPmannose transport and lumenal guanosine diphosphatase activities in Golgi-like vesicles

"Outer-chain" addition of mannose residues to yeast glycoproteins occurs in the Golgi compartment of the cell. Essential steps in this process are thought to include transport of GDPmannose from the cytoplasm into the lumen of Golgi vesicles, transfer of mannose to glycoprotein acceptors, hydrolysis of the resulting GDP to GMP, and return of GMP and inorganic phosphate to the cytoplasm. We report detection and characterization of a GDPmannose transport activity and a GDPase by yeast vesicles. The active transport of GDPmannose as well as the GDPase and another presumed Golgi enzyme, alpha 1,2-mannosyltransferase, are concentrated in a subcellular fraction that can be partially separated, by velocity sucrose gradient centrifugation, from a fraction enriched in an endoplasmic reticulum marker enzyme.

Differential association of rat liver heparan sulfate proteoglycans in membranes of the Golgi apparatus and the plasma membrane

Heparan sulfate proteoglycans (HSPG) of rat liver are associated with the plasma membrane in a hydrophobic intrinsic and a hydrophilic extrinsic form. We were interested in determining whether or not these two forms could be detected in the Golgi apparatus, the subcellular site of addition of oligosaccharides and sulfate to HSPG. In vivo and in vitro radiolabeled HSPG from rat liver Golgi apparatus membranes could only be solubilized with detergents that disrupt the membrane lipid bilayer, suggesting that they are solely associated via hydrophobic interactions. Both forms of HSPG were detected in plasma membranes of rat liver and isolated rat hepatocytes. The detergent-solubilized HSPG bound to octyl-Sepharose columns, whereas the hydrophilic form did not; this latter form, however, was released from the membrane by heparin. The hydrophobic anchor of HSPG in the Golgi and plasma membranes was insensitive to treatment with phosphatidylinositol-specific phospholipase C under conditions in which alkaline phosphatase was sensitive; this suggests that the hydrophobic anchor of HSPG is the core protein itself. Preliminary experiments suggest that the subcellular site of processing of the hydrophobic to the hydrophilic form of HSPG is the plasma membrane. A specific processing activity, probably a protease of the plasma membrane not present in serum or the endoplasmic reticulum membrane, converted hydrophobic HSPG of the Golgi membrane to the hydrophilic form. In addition, pulse-chase experiments with [35S]Na2SO4 in rats demonstrated that at short times, the bulk of the radiolabeled cellular HSPG was in the Golgi apparatus; later on, the bulk of the radioactivity was found in the plasma membrane, the only subcellular site where the hydrophilic form of HSPG was detected.

Mechanism of phosphorylation in the lumen of the Golgi apparatus. Translocation of adenosine 5'-triphosphate into Golgi vesicles from rat liver and mammary gland

The occurrence of phosphorylated secretory proteins such as caseins and vitellogenin and the recent characterization of phosphorylated proteoglycans, in the xylose and protein core, has raised the question of where in the cell and how this phosphorylation occurs. Previous studies have described a casein kinase activity in the lumen of the Golgi apparatus and this organelle as the site of xylose addition to the protein core of proteoglycans. We now report the translocation in vitro of ATP into the lumen of rat liver and mammary gland Golgi vesicles which are sealed and have the same membrane topographical orientation as in vivo. The entire ATP molecule was translocated into the lumen of the Golgi vesicles; this was established by using ATP radiolabeled with tritium in the adenine and gamma-32P. Translocation was temperature dependent and saturable, with an apparent Km of 0.9 microM and Vmax of 58 pmol/mg protein/min. Preliminary evidence suggests that translocation of ATP into the vesicles' lumen is coupled to exit of AMP from the lumen. Following translocation of ATP into the lumen of the vesicles, proteins were phosphorylated.

Reconstitution of Golgi vesicle CMP-sialic acid and adenosine 3'-phosphate 5'-phosphosulfate transport into proteoliposomes

We have previously shown that Golgi apparatus vesicles transport nucleotide sugars and nucleotide sulfate into their lumen. These transport activities are organelle and substrate specific and are characterized by apparent Km for nucleotide derivatives in the low micromolar range. As part of our goal of purifying and characterizing the above transport proteins, we have reconstituted a protein extract from rat liver Golgi membranes into phosphatidylcholine liposomes. The resulting proteoliposomes transport CMP-N-acetylneuraminic acid (CMP-AcNeu) and adenosine 3'-phosphate 5'-phosphosulfate with very similar affinity and inhibition characteristics as intact Golgi vesicles. Sialic acid and sodium sulfate, which are transported only very slowly into the lumen of Golgi vesicles, are transported at low rates by the reconstituted proteoliposomes. Neither rough endoplasmic reticulum-derived vesicles nor proteoliposomes made from proteins of the rough endoplasmic reticulum transport CMP-AcNeu. The above results demonstrate that this reconstituted system can be used for further purification and characterization of nucleotide sugar and nucleotide sulfate translocator proteins. This approach should also be useful to study membrane transport proteins of lysosomes and endosomes.

An intrinsic membrane glycoprotein of the golgi apparatus with O-linked N-acetylglucosamine facing the cytosol

We have recently described the occurrence of integral membrane glycoproteins in rat liver smooth and rough endoplasmic reticulum with O-N-acetylglucosamine facing the cytosolic and luminal sides of the membrane (Abeijon, C., and Hirschberg, C. B. (1988) Proc. Natl. Acad. Sci. U.S.A. 85, 1010-1014). We now report that integral membrane glycoproteins with cytosolic facing O-N-acetylglucosamine also occur in membranes of rat liver Golgi apparatus. This was determined following incubation of vesicles from the Golgi apparatus, which were sealed and of the same membrane topographical orientation as in vivo, with UDP-[14C]galactose and saturating amounts of bovine milk galactosyltransferase. This enzyme does not enter the lumen of the vesicles and specifically catalyzes the addition of galactose, in a beta 1-4 linkage, to terminal N-acetylglucosamine. Under these conditions, galactose was transferred to a glycoprotein of molecular mass of 92 kDa. This protein was insoluble in sodium carbonate, pH 11.5, conditions under which integral membrane proteins remain membrane bound and was insensitive to treatment with peptide:N-glycosidase F. beta Elimination and chromatography showed that radiolabeled galactose was part of a disaccharide which was characterized as Gal beta 1-4GlcNAcitol. This glycoprotein is specific of the Golgi apparatus membrane. Intrinsic membrane glycoproteins with this unusual carbohydrate membrane orientation thus occur in the endoplasmic reticulum and Golgi apparatus of rat liver.

Purification of rat liver N-heparan-sulfate sulfotransferase

N-Heparan-sulfate sulfotransferase catalyzes the transfer of sulfate from 3'-phosphoadenilyl sulfate to the nitrogen of glucosamine in heparan sulfate. This reaction is an obligatory step for subsequent epimerization of D-glucuronic to L-iduronic acid and of O-sulfation of the sugar chains. We have purified this sulfotransferase from rat liver membranes to apparent homogeneity using a combination of conventional and affinity chromatography on DEAE-Sephacel, heparin-agarose, 3',5'-ADP-agarose, wheat germ-Sepharose, and finally a glycerol gradient. The pure enzyme is a glycoprotein with an apparent molecular weight of 97,000. It was enriched in specific activity 65,000-fold over the homogenate. The recovery of activity was 4% of that of the homogenate. Preliminary enzymatic characterization of the purified sulfotransferase indicates a high degree of substrate specificity. Transfer of sulfate occurs to heparan sulfate, N-heparan sulfate, and N-desulfated heparin, but not to N-acetylated heparan sulfate, N-acetylated heparin, chondroitin, chondroitin sulfate, and tyrosine-containing tripeptides.

Intrinsic membrane glycoproteins with cytosol-oriented sugars in the endoplasmic reticulum

We have examined the topography of N-acetylglucosamine-terminating glycoproteins in membranes from rat liver smooth and rough endoplasmic reticulum (SER and RER). It was found that some of these glycoproteins are intrinsic membrane proteins with their sugars facing the cytosolic rather than the luminal side. This conclusion was reached by using vesicles from the SER and RER that were sealed and of the same topographical orientation as in vivo. These vesicles were incubated with UDP-[14C]galactose (which does not enter the vesicles) and saturating amounts of soluble galactosyltransferase from milk, an enzyme that does not penetrate the lumen of the vesicles and that specifically adds galactose to terminal N-acetylglucosamine in a beta 1-4 linkage. Radioactive galactose was mainly transferred to SER proteins of apparent molecular mass 56 and 110 kDa and to a lesser extent to RER and SER proteins of apparent molecular mass 46 and 72 kDa. These proteins are intrinsic membrane proteins, based on the inability of sodium carbonate at pH 11.5 to remove them from the membranes. Studies with peptide N-glycosidase F and chemical beta-elimination showed that the 56-kDa protein of the SER vesicles contained terminal N-acetylglucosamine in an O-linkage to the protein. The above results suggest that some sugars of glycoproteins in the endoplasmic reticulum may attain their final orientation in the membrane by mechanisms yet to be determined.

Subcellular site of synthesis of the N-acetylgalactosamine (alpha 1-0) serine (or threonine) linkage in rat liver

We have studied the subcellular site of synthesis of the GalNAc(alpha-1-0) Ser/Thr linkage in rat liver. The specific and total activities of polypeptide:N-acetylgalactosaminyltransferase (using apomucin as exogenous acceptor) were highly enriched in membrane fractions derived from the Golgi apparatus; virtually no activity was detected in membranes from the rough and smooth endoplasmic reticulum. Vesicles of the above organelles (which were sealed and of the same membrane topographical orientation as in vivo) were able to translocate UDP-GalNAc into their lumen in an assay in vitro; the initial translocation rate into Golgi vesicles was 4-6-fold higher than that into vesicles from the rough and smooth endoplasmic reticulum. Translocation of UDP-GalNAc into Golgi vesicles was temperature dependent and saturable with an apparent Km of 8-10 microM. UDP-GalNAc labeled with different radioisotopes in the uridine and sugar was used to determine that the intact sugar nucleotide was being translocated in a reaction coupled to the exit of luminal UMP. Following translocation of UDP-GalNAc, transfer of GalNAc into endogenous macromolecular acceptors was detected in Golgi vesicles and not in those from the rough and smooth endoplasmic reticulum. The above results together with previous studies on the O-xylosylation of the linkage region of proteoglycans (Nuwayhid, N., Glaser, J.H., Johnson, J.C., Conrad, H.E., Hauser, S.C., and Hirschberg, C.B. (1986) J. Biol. Chem. 261, 12936-12941) strongly suggest that, in rat liver, the bulk of O-glycosylation reactions occur in the Golgi apparatus.